Many proposed and implemented packet classification algorithms trade off JT1emory against lookup-time. Matching algorithms implemented in software cannot keep up with ever-increasing data rates. On the other hand, devices implemented in hardware such as Content Addressable Memory (CAM) have deterministic high lookup rates, but they are expensive in terms of silicon cost and power dissipation. Therefore, a trade-off between hardware and software solutions Le. algorithmic-architectural solutions take advantage of the emerging technologies to provide the required high speed classification without sacrificing the deterministic performance of CAMs. I~ chapter 3 an algorithmic-architectural solution is provided that exploits the geometrical distribution of rules, Hypercuts packet classification algorithm, and CAMs. It works by multi-level cutting ofthe classification space into sub-spaces. It prOVides the deterministic performance of CAMs, support for dynamic updates, and fleXibility for the system designer to trade off the components of the architecture. In chapter 4 Adaptive Rules Cutting (ARC), a heuristic algorithm for packet classification, is proposed. This heuristic algorithm works by selecting the bit positions that divide the classification space into sub-spaces at l-Ievel in a way that reduces the redundancy of rules in cuts and maximizes the equal distribution of rules in cuts. An architecture for packet classification by l-Ievel cutting of the classification space is presented. In chapter 5, architecture for hardware acceleration of session-based IP packet classification is provided and implemented in FPGA. The architecture works by dividing the classification space into sub-spaces at l-Ievel using CRC16. Overall, the thesis provides three new solutions for packet classification based on cutting the classification space into smaller spaces. Supplied by The British Library - 'The world's knowledge'
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:491979 |
| Date | January 2008 |
| Creators | Al-Lagtah, Nasir Mohammed A'mro |
| Publisher | Queen's University Belfast |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.1781 seconds